Bayesian geostatistical modelling of stunting in Rwanda: risk factors and spatially explicit residual stunting burden

Vestine Uwiringiyimana,Sherif Amer,Antonie Veldkamp,Frank Osei

doi:10.1186/s12889-022-12552-y

Vestine Uwiringiyimana, Sherif Amer + Show 2 more

Open Access

https://doi.org/10.1186/s12889-022-12552-y

Copy DOI

Abstract

BackgroundStunting remains a significant public health issue in Rwanda and its prevalence exhibits considerable geographical variation. We apply Bayesian geostatistical modelling to study the spatial pattern of stunting in children less than five years considering anthropometric, socioeconomic and demographic risk factors in Rwanda. In addition, we predict the spatial residuals effects to quantify the burden of stunting not accounted for by our geostatistical model.MethodsWe used the data from the 2015 Rwanda Demographic and Health Survey. We fitted two spatial logistic models with similar structures, only differentiated by the inclusion or exclusion of spatially structured random effects.ResultsThe risk factors of stunting identified in the geostatistical model were being male (OR = 1.32, 95% CI: 1.16, 1.47), lower birthweight (kg) (OR = 0.96, 95% CI: 0.95, 0.97), non-exclusive breastfeeding (OR = 1.24, 95% CI: 1.04, 1.45), occurrence of diarrhoea in the last two weeks (OR = 1.18, 95% CI: 1.02, 1.37), a lower proportion of mothers with overweight (BMI ≥ 25) (OR = 0.82, 95% CI: 0.71, 0.95), a higher proportion of mothers with no or only primary education (OR = 1.14, 95% CI: 0.99, 1.36). Also, a higher probability of living in a house with poor flooring material (OR = 1.22, 95% CI: 1.06, 1.41), reliance on a non-improved water source (OR = 1.13, 95% CI: 1.00, 1.27), and a low wealth index were identified as risk factors of stunting. Mapping of the spatial residuals effects showed that, in particular, the Northern and Western regions, followed by the Southern region of Rwanda, still exhibit a higher risk of stunting even after accounting for all the covariates in the spatial model.ConclusionsFurther studies are needed to identify the still unknown spatially explicit factors associated with higher risk of stunting. Finally, given the spatial heterogeneity of stunting, interventions to reduce stunting should be geographically targeted.

Highlights

Stunting remains a significant public health issue in Rwanda and its prevalence exhibits considerable geographical variation
We modelled the unstructured heterogeneity as normal exchangeable random intercepts uj ∼ N (0, σu2) with uniform prior standard deviation σu ∼ U (0, 10)
30% of the households relied on non-improved sanitation, 78.4% had poor flooring in their house, 84.8% used poor cooking fuels, and 28.4% used a non-improved source for drinking water

Summary

Introduction

Stunting remains a significant public health issue in Rwanda and its prevalence exhibits considerable geographical variation. We apply Bayesian geostatistical modelling to study the spatial pattern of stunting in children less than five years considering anthropometric, socioeconomic and demographic risk factors in Rwanda. Stunting still is a major public health issue in developing countries. Stunting is an indicator of chronic malnutrition in children less than five years. It is defined as height-for-age that is less than two standard deviations. In children, stunting results in decreased motor and cognitive development, impaired immunity, and low education attainment [6]. In adulthood, it leads to lower economic productivity, increases the risk of chronic diseases, and lowers offspring birth weight [6, 7]

Objectives

Methods

Results

Conclusion